[ltp] APM suspend on T23 with 2.6.9

[Michael Perry]

On Mon, 6 Dec 2004 13:54:16 +0200, Marius Gedminas <mgedmin@b4net.lt> wrote:
> Hi,
> 
> I've been using a 2.4.26 kernel on my T23 with great success, but last
> weekend I decided to try out the wonderful world of 2.6.
> 
> I apt-get installed kernel-image.2.6.9-1-686 from Debian unstable,
> added "acpi=off" to kernel parameters in lilo.conf and rebooted.
> 
> Unfortunatelly, APM suspend does not want to work properly: when I press
> Fn+F4 I get a bunch of interesting beeps, the LCD turns off, and the
> disk spins down, but the little moon indicator does not light up, and
> the Ultrabay indicator does not go off.  I also get a message "Disabling
> IRQ 11" from sysklogd.
> 
> After some googling I decided to try /etc/init.d/pcmcia stop before
> suspending.  Suspend now works, however after resuming I still get the
> same "Disabling IRQ 11" message.  Network, USB, PCMCIA and sound stop
> working after that.
> 
> dmesg shows a bunch of messages, the most interesting of which is:
>   irq 11: nobody cared!
>    ...
>    a longish stack trace
>    ...
>   handlers:
>   [<e0acf900>] (snd_intel8x0_interrupt+0x0/0x230 [snd_intel8x0])
>   [<e096c910>] (usb_hcd_irq+0x0/0x70 [usbcore])
>   [<e096c910>] (usb_hcd_irq+0x0/0x70 [usbcore])
>   [<e096c910>] (usb_hcd_irq+0x0/0x70 [usbcore])
>   Disabling IRQ #11
> 
> I tried 2.6.8 as well, but got the same error.
> 
> Any ideas?  Has anyone else experienced similar problems?
> 
> By the way, I'm trying to use APM instead of ACPI because I thought it
> would be simpler.  Also, I remember reading in various places that
> for some people ACPI suspend eats about 10% of battery power per hour,
> while APM suspend needs only 1% per hour.  When I tried ACPI suspend on
> Ubuntu 4.10 Live CD, I noticed that the Ultrabay indicator light does
> not go off.  I suspect that this means Ultrabay power is not
> disconnected -- could this be related to extra battery usage while
> sleeping?  Perhaps I should actually measure ACPI sleep battery usage
> instead of blindly assuming it is abnormal.
> 
> Marius Gedminas
> --
> Veni, Vidi, VISA:  I came, I saw, I did a little shopping.
> 
> 
> 

Just as a matter of "another county heard from", I was never able to
get APM to work on my T23 with the latest bios installed using a 2.6
kernel.  I went to ACPI and initially to a 2.6.8.1 kernel but acpi
failed there and apm caused kernel stack trace errors.  Then I went to
trying acpi with some help from the list archives.  The 2.6.8.1 kernel
would not come back after a suspend and I would get kernel panics and
I had a module or two (like driverloader) which would not suspend.  I
patched that particular kernel with the latest acpi patches from
acpi.sf.net and things started working very well.  I wrote a smallish
script that would allow the lid close events and I added the ibm-acpi
scripts to do suspends using the Fn keys.  Then I went to 2.6.9 and
everything just worked except for lid close events.  I could do
suspend events in a number of ways and it all just worked.

I never got driverloader to behave but OTOH I never had the problems
that some users have had with the USB controller either.  After some
discussion with linuxant support, they told me about a command line
program that would stop the driverloader.  I added that into my lid
close and other suspend event scripts and it all just started working.

For what its worth... I am using the latest bios and power management
updates from ibm.com on the laptop.


-- 
Michael Perry
meperry@gmail.com